[
https://issues.apache.org/jira/browse/IGNITE-20080?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Aleksandr Polovtcev updated IGNITE-20080:
-----------------------------------------
Description:
After Kubernetes-based agents were enabled on TC, unit test builds started to
fail with OOM errors. After having inspected the heap dump, we've discovered
that Kubernetes agents report to have 64 cores in their processors. Since the
number of cores influences the number of threads the Raft component creates, we
ended up with having more than 500 raft-related threads, which consumed all
available memory.
As a quick solution to this problem, I propose to reduce the number of threads
used by the Raft component, at least in tests.
was:
After Kubernetes-based agents were enabled on TC, unit test builds started to
fail with OOM errors. After having inspected the heap dump, I've discovered
that Kubernetes agents report to have 64 cores in their processors. Since the
number of cores influences the number of threads the Raft component creates, we
ended up with having more than 500 raft-related threads, which consumed all
available memory.
As a quick solution to this problem, I propose to reduce the number of threads
used by the Raft component, at least in tests.
> Reduce the number threads used by Raft in tests
> -----------------------------------------------
>
> Key: IGNITE-20080
> URL: https://issues.apache.org/jira/browse/IGNITE-20080
> Project: Ignite
> Issue Type: Task
> Reporter: Aleksandr Polovtcev
> Assignee: Aleksandr Polovtcev
> Priority: Blocker
> Labels: ignite-3
> Time Spent: 10m
> Remaining Estimate: 0h
>
> After Kubernetes-based agents were enabled on TC, unit test builds started to
> fail with OOM errors. After having inspected the heap dump, we've discovered
> that Kubernetes agents report to have 64 cores in their processors. Since the
> number of cores influences the number of threads the Raft component creates,
> we ended up with having more than 500 raft-related threads, which consumed
> all available memory.
> As a quick solution to this problem, I propose to reduce the number of
> threads used by the Raft component, at least in tests.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
